Present Role of Positron Emission Tomography in the Diagnosis and Monitoring of Peripheral Inflammatory Arthritis: A Systematic Review




To determine the current status of positron emission tomography (PET) as a tool for diagnosis and monitoring of peripheral inflammatory arthritis (IA).


For conducting this systematic review, the PubMed (Medline), Embase, and Cochrane Library databases were searched until December 31, 2012. Studies of PET for diagnosis and/or therapy monitoring of peripheral IA were included. Data were summarized qualitatively using best evidence synthesis.


Eighteen articles met our inclusion criteria. The majority of studies were feasibility studies with varying methods applied. All studies demonstrated that PET visualized IA with high sensitivity, corresponding to clinical assessments. PET outcome of clinically active IA also matched that of ultrasound and magnetic resonance imaging. PET differentiates from other modalities by (quantitative) imaging of molecular sites in the synovium. The first studies reporting on the potential clinical applications of PET to image subclinical synovitis in preclinical RA and during therapy have been published. The results are promising, but the number and study populations of these studies are still limited.


Thus far, a limited number of PET studies addressing IA imaging have been published. The PET modality seems to offer highly sensitive and potentially specific imaging of IA at the (quantitative) molecular level. Clinical application studies for early diagnostics and therapy monitoring are arising, but these topics should be further explored in future studies with larger cohorts. For integration in clinical practice, aspects such as radiation burden and cost-effectiveness should also be taken into account.


Positron emission tomography (PET) is a sensitive, quantitative, and noninvasive imaging technique that visualizes functional tissue changes by targeting binding sites ([1]). PET provides molecular data as opposed to anatomic and functional data obtained by ultrasonography (US) and magnetic resonance imaging (MRI). In addition, compared with other nuclear imaging techniques such as conventional scintigraphy, a great advantage of PET is the high level of sensitivity in 3-dimensional mode and its ability to quantify tracer uptake accurately and reproducibly, enabling monitoring of disease activity and therapeutic effects ([2]).

PET was developed in the mid-1970s after the introduction of fluorodeoxyglucose (FDG) labeled with fluorine-18 (18F-FDG) ([3]). Three decades later, PET–computed tomography (CT) combinations were introduced for more precise anatomic localization of the PET signal ([4]).18F-FDG is taken up in metabolically active tissues and phosphorylation prevents the glucose from being released from the cell once it has been absorbed. 18F-FDG PET is extensively used in the field of oncology for diagnosis, staging, and followup ([5]), but its application for rheumatologic diseases is relatively new ([6]).

The potential of 18F-FDG PET to visualize inflammatory lesions was first noticed on oncologic PET scans revealing uptake in noncancer/benign and metabolically active lesions ([7]). Subsequently, increasing numbers of studies reported the feasibility to image inflammatory arthritis (IA) by PET. Advanced imaging techniques such as PET are required to fulfill the clinical needs of early diagnosis of IA and optimization of therapeutic efficacy because synovitis is not shown on conventional radiographs.

In this systematic review, the current role of PET as a tool for the diagnosis and therapy monitoring of IA is discussed. We reviewed PET studies of IA, consisting of rheumatoid arthritis (RA), juvenile idiopathic arthritis (JIA), ankylosing spondylitis (AS), psoriatic arthritis (PsA), and reactive arthritis (ReA).

Box 1. Significance & Innovations

  • Positron emission tomography (PET) allows for sensitive, noninvasive imaging of inflammatory arthritis (IA) of the whole body in 1 imaging session. Specificity may be achieved by application of tracers that bind to molecular inflammatory targets in synovial tissue. PET has advantages over scintigraphy for the imaging of arthritis due to a higher sensitivity for the detection of low-grade inflammation, 3-dimensional imaging, and the potential to quantify tracer uptake.
  • Because of costs and radiation burden, PET will not have clinical value for the diagnostics of clinically accessible arthritis. However, it has the potential (as indicated by recent study data) to address the clinical needs of detection of early (preclinical) IA disease activity and monitor or ultimately predict the therapeutic outcome (in particular of expensive biologic agents) through sensitive imaging of subclinical synovitis and quantification of the PET data. Studies with large cohorts, uniform scanning protocols, and comparable outcome measures are warranted to further explore and potentially validate PET for these clinical applications.
  • The novel characteristics of PET in imaging at the molecular level of synovial tissue distinguish this technique from anatomic and functional techniques such as magnetic resonance imaging (MRI) and ultrasound (US). Feasibility studies have demonstrated comparable sensitivity of the different techniques to image clinically active IA. However, head-to-head comparisons of PET to MRI and/or US (or by application of hybrid imaging PET-MRI) for detection of subclinical synovitis are still lacking. Specificity may be an advantage of PET but needs to be addressed in future studies using 2 or more imaging modalities.


Eligibility criteria

Data for this review were extracted from the collection of primary diagnostic studies reporting the use of PET, with any kind of tracer, for the diagnosis and/or therapy monitoring of peripheral IA in patients. The review included patients with RA, JIA, AS, PsA, or ReA with active peripheral arthritis on the basis of a clinical examination. To investigate the value of PET in the early diagnosis of IA, studies with patients in preclinical disease stages were included as well. Review articles, case reports, animal studies, scintigraphic studies without PET, and studies of patients with forms of arthritis other than those stated above (e.g., septic arthritis and spondyloarthritis) were excluded.

Information sources and search methods

Studies were identified by searching the PubMed (Medline), Embase, and Cochrane Library databases (trials) up to December 31, 2012. The search strategies for the different databases are shown in Supplementary Appendix A (available in the online version of this article at Because PET research in rheumatology is a relatively new topic, we performed a broad search on PET (imaging technique) in IA (disease group) without narrowing the search by a (third) outcome measure (like diagnosis or therapy monitoring). In line with this, no limits were applied for the year of publication and language (to avoid initial publication bias). Related terms in the articles and reference lists of retrieved articles were searched as well.

Study selection

Two authors (STGB and YYJG) independently screened the titles, keywords, and abstracts of retrieved studies for relevance. After consensus, the full-text versions of studies that met the eligibility criteria and of studies for which eligibility could not be determined from initial screening of the title and/or abstract were acquired. After independently assessing the full-text articles, the final inclusion was determined by consensus. In the case of disagreement, 2 authors (AEV and CJvdL) were consulted for a final decision.

Data extraction and quality assessment

To achieve uniformity, we developed 4 data extraction sheets for the 1) study population characteristics, 2) imaging characteristics (see Supplementary Appendix B, available in the online version of this article at, 3) study outcomes, and 4) therapy monitoring characteristics. One reviewer (STGB) extracted the data from the included studies, independently controlled by a second reviewer (YYJG).


Study selection and inclusion

The initial search of the Pubmed (Medline), Embase, and Cochrane Library databases retrieved 2,412 articles referring to PET research in IA. Figure 1 shows the methods resulting in a final inclusion of 18 original articles.

Figure 1.

Flowchart of literature selection. Adapted, with permission, from ref. [53]. For more information, visit

Study characteristics

All assessed articles were published in English. From the first study onward (1995), in total, 18 studies including a total of 276 IA patients (175 females, 65 males, and 36 unknown) predominantly diagnosed with RA (79%) according to the American College of Rheumatology 1987 criteria were performed ([8]) (Tables 1-4).

Table 1. Study characteristics of active IA*
Author, year (ref)No. patientsAge, yearsaJoint(s)TracerPET variableComparative variable
  1. IA = inflammatory arthritis; PET = positron emission tomography; RA = rheumatoid arthritis; PsA = psoriatic arthritis; FDG = fluorodeoxyglucose; RUV = regional uptake value defined as fixed circular regions of interest of 16 mm drawn on top of FDG uptake; TUV = total uptake value defined as the sum of standardized uptake value (SUV) in individual axial images; MCP = metacarpophalangeal; MTP = metatarsophalangeal; VA = visual assessment of tracer uptake (pattern); CSUV = cumulative SUV; SJC = swollen joint count; TJC = tender joint count; DAS = Disease Activity Score; SDAI = Simplified Disease Activity Index; PGA = patient global assessment; phyGA = physician global assessment; SUVmax = maximum SUV value; PJC = pain joint count; JIA = juvenile idiopathic arthritis; CRP = C-reactive protein; ESR = erythrocyte sedimentation rate; MMP-3 = matrix metalloproteinase 3; pp = per person; DAS28 = DAS in 28 joints; DAS28-CRP = DAS28 using the C-reactive protein level ([14]); RF = rheumatoid factor; SSA = seronegative spondyloarthropathy; AS = ankylosing spondylitis; nsSSA = nonspecific SSA; RADAI = Rheumatoid Arthritis Disease Activity Index; lbm = lean body mass; SUV J/C ratio = SUV from a joint divided by the SUV of a control region; OA = osteoarthritis; FM = fibromyalgia; SpA = spondyloarthritis; ReA = reactive arthritis; JRA = juvenile rheumatoid arthritis; Ki = influx constant; uSpA = undefined SpA; PBR = peripheral benzodiazepine receptor.
  2. aValues are the mean, mean ± SD, range, mean (range), or mean ± SD (range).
  3. bRA following the American College of Rheumatology 1987 criteria.
Palmer et al, 1995 ([10])9 RA, 3 PsAb35–75Wrist18F-FDGRUV, TUVPain, tenderness, swelling
Beckers et al, 2004 ([11])21 RAb48 (34–69)Knee (21 patients), wrist/MCP (13 patients), ankle/first MTP (8 patients)18F-FDGVA, CSUVSJC, TJC, disease duration, DAS, SDAI, PGA and phyGA
Kubota et al, 2009 ([12])18 RAb67 ± 11 (47–86)Multiple (13 pp)18F-FDGTotal SUVmax, VAClinically positive/negative joints, PJC, SJC
Tateishi et al, 2010 ([13])28 JIA5.4 ± 4 (1–16)560 joints18F-FDGSUVmaxSwelling, tenderness, CRP level, neutrophil count, ESR, MMP-3
Okamura et al, 2012 ([14])22 RAb57 (20–74)Multiple (12 pp)18F-FDGMean SUVmaxDAS28, DAS28-CRP, TJC, SJC, CRP level, ESR, MMP-3, RF
Vijayant et al, 2012 ([15])17 RAb7 AS3 PsA1 nsSSA40 (27–60)29 (17–40)42 (34–53)19Multiple (21 pp)18F-FDGVAClinical diagnosis of RA, SSA, PsA
Goerres et al, 2006 ([16])7 RAb50 ± 18 (24–63)Multiple (28 pp)18F-FDGVADAS, RADAI
Roivainen et al, 2013 ([17])17 RAb51 ± 14Multiple (54 pp)18F-FDGVASJC, TJC
Beckers et al, 2006 ([18])16 RAb48 (34–63)Knee18F-FDGSUVlbmCRP level, MMP-3, tender/swollen joints
Elzinga et al, 2007 ([19])14 RAb6 OA5 FM57 ± 755 ± 1132 ± 11Hands and wrists18F-FDGSUV J/C ratioSJC, DASClinical diagnosis of RA, OA and FM
Taniguchi et al, 2010 ([20])7 RA2 AS1 PsA1 ReA63 ± 1048 and 507263Multiple18F-FDGSUVmax, VAClinical diagnosis of SpA and RA
Okabe et al, 2011 ([21])30 RA2 AS63 ± 15Multiple (19 pp)18F-FDGVA, SUVmax of jointsClinical diagnosis of IA and collagen vascular disease–associated arthritis
Danfors et al, 1997 ([22])3 RA1 JRA1 PsA592436Knee11C-d-deprenylSUVIA and healthy controls
Roivainen et al, 2003 ([23])2 RAb1 AS6 unspecified1 uSpA25 and 604820–4835Knee (9)Ankle (1)18F-FDG11C-cholineSUV, KiSJC, TJC
Van der Laken et al, 2008 ([24])10 RAb54 ± 8Knees11C-(R)-PK11195SUV J/C ratioSemiquantitative scores of synovial swelling and PBR staining of synovial tissue
Table 2. Study characteristics comparing PET with other imaging modalities*
Author, year (ref)ModalityPET variableComparative variable
  1. For the complete patient characteristics of the studies in Table 2, see Table 1. PET = positron emission tomography; MRI = magnetic resonance imaging; RUV = regional uptake value defined as fixed circular regions of interest of 16 mm drawn on top of 18F-fluorodeoxyglucose (18F-FDG) uptake; TUV = total uptake value defined as the sum of the standardized uptake value (SUV) in individual axial images; VEP = volume of enhancing pannus; Δ = pre- versus postintervention; Ga-67 = Gallium-67; VA = visual assessment; US = ultrasonography; CSUV = cumulative SUV; lbm = lean body mass; RE = relative enhancement; SE = static enhancement.
Palmer et al, 1995 ([10])PET, MRIRUV, TUV ΔRUV, ΔTUVMRI (VEP in mm3) ΔMRI (VEP in mm3)
Roivainen et al, 2003 ([23])PET, MRISUV of 18F-FDG and 11C-cholineMRI: volume of synovium (in cm3) and enhancement integral
Taniguchi et al, 2010 ([20])PET, MRI, Ga-67 scintigraphyVAMRI and Ga-67 scintigraphy
Beckers et al, 2004 ([11])PET, USVA, CSUV(Cumulative) synovial thickness and Doppler score on US
Beckers et al, 2006 ([18])PET, US, MRISUVlbm, ΔSUVlbmUS/ΔUS synovial thickness (mm) MRI/ΔMRI RE (55, 30, and 15 seconds) and SE (15 minutes)
Table 3. Study characteristics reporting subclinical disease activity*
Author, year (ref.)No. patientsaAge, yearsJointsTracerPET variableComparative variable
  1. PET = positron emission tomography; RA = rheumatoid arthritis; FDG = fluorodeoxyglucose; VA = visual assessment; pp = per person; SUV = standardized uptake value; ACPA = anti–citrullinated protein antibody–positive arthralgia patients.
  2. aValues are the mean (range), mean ± SD, or mean ± SD (range).
  3. bRA following the American College of Rheumatology 1987 criteria.
Beckers et al, 2004 ([11])21 RAb48 (34–69)21 joints (see Table 1)18F-FDGVAClinical joint assessment
Elzinga et al, 2007 ([19])14 RAb57 ± 7Hands and wrists18F-FDGVAClinical joint assessment
Kubota et al, 2009 ([12])18 RAb67 ± 11 (47–86)Multiple (13 pp)18F-FDGSUVAssessment of clinical remission
Goerres et al, 2006 ([16])7 RAb50 ± 18 (24–63)Multiple (28 pp)18F-FDGVAClinical total joint score
Van der Laken, 2008 ([24])10 RAb54 ± 8Knees11C-(R)-PK11195SUV ratio arthritis/ contralateral jointRA patients and controls
Gent et al, 2012 ([36])29 ACPA3 RAb44 ± 1249 ± 10Hands and wrists11C-(R)-PK11195VAClinical assessment at baseline and followup
Table 4. Study characteristics reporting treatment monitoring*
Author, year (ref)No. patientsAge, yearsaJointsTracerPET variableComparative variablebTherapy
  1. PET = positron emission tomography; RA = rheumatoid arthritis; NA = not available; FDG = fluorodeoxyglucose; RUV = regional uptake value defined as fixed circular regions of interest of 16 mm drawn on top of 18F-FDG uptake; TUV = total uptake value defined as the sum of standardized uptake value (SUV) in individual axial images; Paulus Index = analysis of treatment response ([39]); NSAID = nonsteroidal antiinflammatory drug; MTX = methotrexate; PsA = psoriatic arthritis; Δ = pre- versus postintervention; JRA = juvenile rheumatoid arthritis; pp = per person; VA = visual assessment; EULAR = European League Against Rheumatism; SUVmax = maximum SUV value; TJC = tender joint count; SJC = swollen joint count; DAS28 = Disease Activity Score in 28 joints; DAS28-CRP = DAS28 using the C-reactive protein level ([14]); ESR = erythrocyte sedimentation rate; MMP-3 = matrix metalloproteinase 3; RF = rheumatoid factor; lbm = lean body mass; anti-TNFα = anti–tumor necrosis factor α; DAS28-ESR = DAS28 using the ESR; VAS = visual analog scale; ROM = range of motion; M-HAQ = modified Health Assessment Questionnaire; MCP = metacarpophalangeal.
  2. aValues are the mean, range, mean (range), mean ± SD, or mean ± SD (range).
  3. bMost patients were investigated within 1 week of physical examination for clinical response assessment.
  4. cRA following the American College of Rheumatology 1987 criteria.
Palmer et al, 1995 ([10])9 RAc3 PsANA35–75Wrist18F-FDGRUV, TUVPaulus IndexNSAID (piroxicam), prednisone, MTX
Danfors et al, 1997 ([22])3 RA1 JRA1 PsA592436Knee11C-d-deprenylΔSUVClinical improvement (no detailed information available)Steroid injections L-deprenyl, synovectomy
Goerres et al, 2006 ([16])7 RAc50 ± 18 (24–63)Multiple (28 pp)18F-FDGVA (score 0–4)EULAR response criteriaInfliximab
Vijayant et al, 2012 ([15])9 RAcNAMultiple (21 pp)18F-FDG% change in SUVmax, VA% change in TJC, SJCSUVmaxMTX, hydroxychloroquine, prednisolone
Okamura et al, 2012 ([14])22 RAc57 (20–74)Multiple (12 pp)18F-FDGΔmean SUVmax, SUVmaxΔDAS28, ΔDAS28-CRP, ΔTJC, ΔSJC, ΔCRP level, ΔESR, ΔMMP-3, ΔRFEtanercept, infliximab
Beckers et al, 2006 ([18])16 RAc48 (34–63)Knee18F-FDGΔSUVlbmΔCRP level/ΔMMP-3Anti-TNFα therapy
Roivainen et al, 2013 ([17])17 RAc51 ± 14Multiple (54 pp)18F-FDGSUVΔSUVmaxΔSUVmax (at 2 and 4 weeks)SJC, TJC, ESR, CRP levelΔCRP level/ΔESRΔDAS28-ESR (at week 12)MTX, sulfasalazine, hydroxychloroquine, prednisolone
Sato et al, 2009 ([37])6 RAc61 ± 12 (41–75)Knee18F-FDGSUVmaxVolume SUV >1 in cm3VAS, ROM, faceScale for mood, M-HAQ, ESR, CRP levelAcupuncture
Elzinga et al, 2011 ([38])16 RAc53 ± 10MCP and wrists18F-FDGΔmean SUV (0–2 weeks)DAS 14 weeks/ 22 weeksInfliximab

The research questions comprised the feasibility of PET to assess the disease activity of IA (n = 8), therapy monitoring (n = 9), and detection of preclinical disease activity (n = 1). PET was the only imaging modality in 6 studies, and 12 studies used PET (PET-CT) next to other imaging modalities like MRI and US. Most studies used 18F-FDG as the PET tracer (15 of 18).

All studies evaluated PET data visually, but most studies also included a semiquantitative analysis (Tables 1-4). For visual assessment, a PET scan was classified as positive if tracer uptake was present in at least 1 joint. Semiquantitative tracer uptake is usually reported as the standardized uptake value (SUV) ([9]), representing the ratio of tissue radioactivity concentration (measured from the PET scan) to the injected dose typically normalized for body weight (or another measure of volume of distribution, e.g., lean body mass). The tissue radioactivity concentration can be shown as the mean (SUVmean) or maximum value (SUVmax) as measured within a volume or region of interest. Sometimes, other calculations are derived from SUV, like a regional uptake value (RUV) or total uptake value (TUV), defined as the SUV in fixed planar regions of interest or the sum of the SUV in individual axial images, respectively ([10]). Semiquantitative data can also be obtained by the calculation of joint to control ratios. Regions of interest are then drawn on joints and (fixed) background areas.

Potential of PET to visualize clinically active IA

The potential of PET to visualize arthritis has been tested in several studies of patients with clinically active IA (Table 1). Studies by Palmer et al and Polisson et al were the first to show that 18F-FDG PET could visualize clinical arthritis by targeting glucose metabolism (in 1995). The PET images showed locally increased 18F-FDG uptake in clinically inflamed wrist joints of RA patients ([10, 25]). In addition, the semiquantitative uptake of 18F-FDG (SUV) correlated with clinical evaluations like joint swelling (TUV: r = 0.53, P = 00.2 and RUV: r = 0.74, P = 1 × 10−6) ([10]). This association of SUV values of 18F-FDG with clinical findings in IA cohorts was confirmed by other studies including children with JIA ([11-15]). (Cumulative) 18F-FDG joint uptake on PET also correlated with validated clinical methods for the assessment of disease activity of RA, for example, the disease activity score in 28 joints (DAS28; up to r = 0.78, P = 0.002) ([11, 16, 26]) and the Simplified Disease Activity Index (r = 0.90, P < 0.0001) ([27]). Instead of PET scan evaluation by quantification of the tracer uptake in joints (SUVs), more simplified PET scoring tests have also been investigated. Three studies ([12, 15, 16]) found that simple visual scoring systems based on 18F-FDG joint uptake (i.e., grading 0–4 [16]) also corresponded to clinical symptoms of arthritis.

Different studies reported concordant PET uptake in >90% of clinically active joints ([17, 23]). However, while most studies included small populations, 3 larger 18F-FDG PET studies provided some detailed data to make estimations on the diagnostic performance of PET for IA imaging ([11, 18, 19]). Derived from these studies, the sensitivity for detecting IA, with clinical assessment as the reference, varied between 56% and 77%.

18F-FDG PET seems to be highly specific in the discrimination of arthritic joints from healthy joints ([11]) (Figure 2). However, it cannot distinguish reliably between RA and osteoarthritis. In a comparative study, more joints were PET positive in RA than in osteoarthritis (29% and 6%, respectively), but at the joint level, absolute tracer uptake did not allow a distinction between the two ([19]). The uptake of 18F-FDG in degenerative joints has also been described in other studies ([28, 29]). Although not very specific at the joint level, at the patient level, distribution of 18F-FDG uptake in different joints shown on a whole-body scan may also help to differentiate one inflammatory rheumatic disease from the another ([15, 20, 21]). For example, 18F-FDG uptake in RA patients has been typically observed in the synovial tissue of (sub)clinically inflamed joints, whereas in spondyloarthritis (SpA) patients, a typical tracer uptake pattern at the entheses, reflecting enthesitis, has been observed ([15, 20]).

Figure 2.

18F-FDG PET images of a healthy control subject and a patient with rheumatoid arthritis with active disease. A, 3-dimensional projection image of tracer distribution in a healthy knee. B, Distribution in a healthy hand and wrist. C, A rheumatoid knee. D, A rheumatoid hand and wrist. Reproduced, with permission, from ref. [11].

Radiopharmaceuticals other than 18F-FDG may also be interesting for imaging IA. Danfors et al found that the tracer 11C-d-deprenyl accumulated in arthritic ankle joints ([22]). The definite mechanism underlying 11C-d-deprenyl uptake in inflamed tissues has not been elucidated yet. The relationship of 11C-d-deprenyl with inflammation is strengthened by the observation that corticosteroid injections block 11C-d-deprenyl retention in inflammatory tissue ([30]).

To improve the specificity of PET tracer uptake in IA joints versus non-IA joints, specific tracers targeting molecular pathways expressed in IA synovitis are needed. A few studies have investigated the feasibility of potentially specific tracers in RA patients (see Supplementary Appendix C, available in the online version of this article at

Because rheumatoid synovitis is characterized by proliferative changes of synovial tissue, methyl-11C-choline (choline is a precursor for synthesis of phospholipids) has been explored as a tracer ([31]). In 10 patients with inflammatory joint disease (among whom 2 had RA), the tracer was compared with 18F-FDG ([23]). PET images showed fast and high accumulation of both 18F-FDG and 11C-choline at sites of synovitis (mean ± SD SUVs of 1.9 ± 0.9 and 1.5 ± 0.9, respectively). Although 11C-choline uptake rates were 8-fold faster compared with 18F-FDG (mean ± SD kinetic influx constant of 0.048 ± 0.042 minute−1versus 0.006 ± 0.003 minute−1), this also implied a higher chance of dependency on blood flow effects.

Another potential target to image arthritis is the macrophage, because macrophages infiltrate in synovium as RA develops and continue to play an important role in later phases. Targeting peripheral benzodiazepine receptors, particularly expressed on activated macrophages with 11C-PK11195, has been successful to visualize inflammatory lesions in the brain as multiple sclerosis and vasculitis ([24, 32]). In established RA patients with clinical arthritis of one or both knee joints, 11C-(R)-PK11195 clearly accumulated in clinically inflamed knee joints, with mean ± SD SUV ratios of 3.2 ± 0.9, 2.3 ± 0.6, and 1.7 ± 0.4 for severe, mild, and absent clinical signs of arthritis, respectively. The SUV ratios were significantly different between groups (P < 0.05). In addition, PET results highly corresponded with clinical swelling and macrophage infiltration in synovial tissue (immunohistochemistry) ([24]).

Comparison of PET with MRI and US in clinically active IA and treatment monitoring

Serial imaging of IA patients with PET and MRI has been reported in 4 of the included feasibility studies in clinically active RA ([10, 18, 20, 23]) (Table 2). Both baseline 18F-FDG uptake and pannus volume as measured with MRI, as well as their changes after treatment, correlated strongly (r = 0.86, P = 0.0001 and r = 0.91, P = 0.0002, respectively) ([10, 25]). In line with this, uptake of both 11C-choline and 18F-FDG in synovium (in SUVs) corresponded with the volume of synovium on MRI, with the highest correlation coefficient for 11C-choline uptake ([23]), suggesting that uptake of 11C-choline in proliferative synovial tissue was more concordant with synovial volume on MRI than focal metabolic uptake of 18F-FDG (18F-FDG: r = 0.86, P = 0.002 and 11C-choline: r = 0.95, P < 0.0001).

For imaging of entheses in SpA, 18F-FDG PET-CT seemed to be more sensitive than MRI at lumbar spinal processes (87.5% [7 of 8] versus 50% [3 of 6]) and at ischial tuberosities (75% [6 of 8] versus 0% [0 of 5]) ([20]). Furthermore, 18F-FDG PET results of RA patients showed correlations with US. Cumulative 18F-FDG PET positivity at the group level (n = 365 joints) was significantly associated with swelling (odds ratio [OR] 4.8), tenderness (OR 8.6), and US positivity (OR 11.7) ([11]). Interestingly, the prevalence of 18F-FDG PET–positive joints (>75%) was significantly higher than the prevalence of US-positive joints (56%). 18F-FDG PET was positive in 86% of the US-positive joints and was significantly more concordant in joints with a positive power Doppler (PD) signal (i.e., PET was positive in 44 of 45 PD-positive joints [96%]). Such a signal is found in joints with hypervascularization, which is supposed to be a characteristic of active RA synovitis. The mean SUV of PD-positive joints on PET was also significantly higher than in PD-negative joints. In addition, a significant correlation between 18F-FDG uptake and synovial thickness as measured by US was found for nearly all joints, except for the first metatarsophalangeal joints (all 365 joints: r = 0.56, P = 0.0001). This relationship was stronger for larger joints, such as the knees and ankles, than for smaller joints, such as the metacarpophalangeal joints, proximal interphalangeal joints, and wrists. This phenomenon may be explained by the technical properties of US and PET. The detection of synovitis with US is more difficult at sites that are not easily accessible for the probe in different planes, such as the metacarpophalangeal joints (3 and 4) and the intercarpal compartments of the wrist ([33, 34]). Besides, US does not penetrate bone. Conversely, PET has a limited spatial resolution (typically 5–6 mm) that limits the detection of small inflammatory lesions.

At present, 1 study included all 3 imaging modalities (PET, MRI, and US) in a group of IA patients ([18]). The results of this study showed that a positive joint on PET was significantly associated with a positive sign of arthritis on MRI and US (i.e., of the 11 PET-positive knees, 10 were also MRI positive and 10 were US positive). Moreover, in RA knee joints, changes in SUV over a period of 4 weeks of treatment also correlated with changes in MRI parameters, such as relative enhancement (r = 0.60, P = 0.015), rate of early enhancement (r = 0.58, P = 0.0177), and static enhancement (r = 0.52, P = 0.0396) ([35]), but not with changes in synovial thickness as measured with US ([18]).

Subclinical IA disease activity

The visualization of subclinical synovitis in IA patients by advanced imaging modalities has been reported in several studies (Table 3). In 2 18F-FDG PET studies, at the joint level, an average of 10% of the evaluated joints of RA patients showed signs of tracer uptake despite an absence of clinical signs of inflammation ([11, 19]). In addition, 2 other 18F-FDG PET studies reported that the cumulative and mean number PET-positive joints were (significantly) higher than the number of joints with clinical symptoms of synovitis (mean ± SD 46.4 ± 16.5 cumulative PET-positive joints and 41.7 ± 16.2 joints with clinical symptoms of synovitis [n = 7] and mean ± SD 6.2 ± 3.3 PET-positive joints and 3.1 ± 2.7 joints with clinical symptoms of synovitis [n = 17; P = 0.0002]) ([12, 16]). Moreover, by targeting macrophages using 11C-(R)-PK11195 PET, the SUV ratios in 5 clinically uninflamed knee joints (mean ± SD 1.7 ± 0.4) of RA patients were significantly higher than those in uninflamed knee joints of control subjects (mean ± SD 1.1 ± 0.1; P < 0.05) ([24]). Interestingly, one 18F-FDG PET study in RA included 4 remission patients. Although tracer joint uptake was lower than in those patients with clinically active disease (total mean ± SD SUVmax 22.0 ± 1.8 versus 33.4 ± 12.1; P < 0.001), all 4 remission patients had at least 1 PET-positive joint ([12]).

Although a false-positive PET signal in these asymptomatic joints cannot be excluded, a recent prospective study supports the relevance of such observations. All of the 4 anti–citrullinated protein antibody–positive arthralgia patients without clinical signs of arthritis with a positive 11C-(R)-PK11195 PET at baseline developed IA within 1 year of clinical followup, whereas the majority (20 of 25 patients) of those with a negative 11C-(R)-PK11195 PET at baseline did not develop any IA in a 2-year followup period ([36]). Only 2 of 5 patients with a negative PET scan of the hands developed subtle arthritis in 1 or 2 hand joints while the others developed arthritis outside the field of view of the PET scan. These data support detection of subclinical IA in the very early phase of developing RA by PET and macrophage targeting (Figure 3).

Figure 3.

Representative 11C-(R)-PK11195 images of the hands and wrists in A, an anti–citrullinated protein antibody–positive individual with arthralgia, B, a rheumatoid arthritis patient with clinical arthritis in various hand joints, and C, a healthy volunteer without joint uptake. Arrows indicate examples of joints with elevated tracer uptake. Note the background uptake of the tracer in intrinsic hand muscles and bone marrow and around the nails in all images. R = right hand. Reproduced, with permission, from ref. [36].

Treatment monitoring and prediction

The characteristics of PET to quantify tracer uptake at the molecular level potentially allow for treatment monitoring. This application has been addressed in several studies in IA (Table 4). A decline (up to 58%) of both 18F-FDG and 11C-d-deprenyl was noticed in arthritic joints after (antirheumatic) drug interventions ([10, 17, 22]). Furthermore, 18F-FDG PET data, both quantitative and visual, were in agreement, with clinical changes of disease activity (DAS28) during anti–tumor necrosis factor treatment ([14, 16]). In 22 RA patients, the mean ± SD of the cumulative SUVmax of 12 joints per patient significantly decreased from 2.14 ± 0.55 to 1.66 ± 0.48 (P < 0.005) and the mean ± SD DAS28 score decreased significantly from 5.29 ± 1.01 to 3.81 ± 0.86 (P < 0.001); these independent changes correlated significantly (r = 0.538, P = 0.01) ([14]). A simple visual PET scoring system for response assessment, investigated in 7 RA patients, also showed a significant correlation between a decrease of PET signal (mean ± SD 46.4 ± 16.5 to 41.7 ± 16.2) and DAS28 score (mean ± SD 5.6 ± 1.2 to 3.3 ± 1.2; P = 0.04) ([16]). It should be noted that the determination of tracer uptake by visual assessment is less accurate than (semi)quantification of the uptake. An example of this was shown in a study by Vijayant et al; in this therapy monitoring study, the authors found that 18F-FDG uptake remained visually high despite decreases in SUVmax ([15]).

Changes in quantitative PET data also seemed to correlate with changes in serum inflammation levels, such as C-reactive protein (CRP) after 4 weeks (r = 0.51, P = 0.004) and 12 weeks (r = 0.90, P = 0.002) of treatment ([17, 18]). Conversely, in treatments with a lack of change in serum inflammation levels such as acupuncture, PET corresponded to an absence of significant SUV changes ([37]).

Besides treatment monitoring, PET might also allow for the prediction of treatment efficacy early in the course of treatment, or ultimately at baseline. Two studies compared 18F-FDG PET data of RA patients before and a few weeks after treatment ([17, 38]). In 16 RA patients, a significant correlation was found between early changes (0–2 weeks) of 18F-FDG uptake (mean SUV) in the hands and wrists and DAS28 score at 14 weeks (r = 0.62, P = 0.05) and 22 weeks (r = 0.65, P = 0.01). In addition, in 17 RA patients, the percentage of 18F-FDG PET changes of multiple joints (also including the hands and wrists) between baseline and 2 or 4 weeks corresponded with the percentage of changes of clinical disease activity (DAS28 using the erythrocyte sedimentation rate [ESR]) at 12 weeks (r = 0.58, P = 0.014 at 2 weeks, data not available at 4 weeks) ([17]). In contrast, no correlation was found between changes in ESR, CRP level, and DAS28 between 0–2 weeks of treatment and clinical outcome at 14 weeks ([38]), suggesting that 18F-FDG has a stronger prognostic value for treatment response than currently available parameters in the early phase of treatment.


This review included 18 primary studies that reported the results of PET research in patients with peripheral IA, primarily RA. The proof of concept of sensitive semiquantitative imaging of IA by PET was demonstrated in these studies. Because of high costs and radiation burden, PET will not be useful in clinical practice for the imaging of clinically active IA that can be easily determined by clinical assessments. However, because the spectrum of IA is changing toward diagnosis in the preclinical phase of disease, advanced imaging detecting subclinical disease activity is required. Moreover, the clinical need to increase the efficacy of expensive biologic agents warrants sensitive and quantitative monitoring of therapy. PET studies that address these needs are now arising. The first results of such studies are promising and may suggest a role for PET in early diagnostics and for monitoring or ultimately the prediction of therapeutic efficacy, eventually contributing to individualized therapy. Likewise, PET has already been implemented in the clinical practice of oncology for the diagnosis, staging, and therapy monitoring of the disease ([5]). Finally, positive correlations between simple visual PET scoring and clinical assessments may enhance implementation of PET analysis in clinical practice.

Although the yield of information derived from this review is valuable, this review has its limitations. The number of PET studies investigating imaging of peripheral IA is still limited. Most studies had a proof-of-concept design focused on the feasibility of imaging of clinically active IA (mostly focused on RA) with small sample sizes and great heterogeneity in study design, patient selection, and analysis. The latter implies that pooling of data was not possible and consequently the results were presented in a descriptive manner.

Most PET studies in this review (15 of 18 studies) used 18F-FDG as the tracer. Although this tracer performs well in the imaging of IA, it is nonspecific and accumulates in any joint that is metabolically active, as also noticed in osteoarthritis and other systemic (infection) diseases ([19, 40-44]). The application of specific tracers targeting different processes in the inflammatory cascade of IA ([45]) may contribute to increased specificity ([22-24, 36]). Besides the application of specific tracers, in vivo biodistribution of the tracer (e.g., predominantly uptake in the synovium in RA versus uptake in the entheses in SpA) may also contribute to the differentiation of inflammatory joint diseases ([12, 20]).

The advantage of the technique of PET to image and quantify pathophysiology at the molecular level as well as the option to apply specific tracers makes PET a highly sensitive and potentially specific modality for imaging subclinical IA. With the introduction of the hybrid technique PET-CT, the molecular information can be localized more precisely by using (low-dose) CT scans for anatomic delineation of the PET signal. In particular, with regard to molecular quantification and the aspect of specificity, PET may have advantages over anatomic/functional imaging techniques such as MRI and US. These latter techniques have suboptimal quantification in a longitudinal setting ([46]) for the detection of subtle early or residual synovitis activity, which is relevant for early diagnostics and therapy monitoring. The other advantages of PET are whole-body imaging in 1 session within 20–30 minutes (or even shorter for particular tracers as derived from kinetic analyses) and independency of operator skills.

The disadvantages of PET are the use of radioactivity, the poor spatial resolution (and hence a major impediment to quantification), and the costs and variable availability of PET facilities. The use of radioisotopes can lead to an effective dose ([47]) of ∼7 mSv when 370 MBq of the tracer 18F-FDG is used. A low-dose whole-body attenuation CT would add ∼3–5 mSv, leading to a total PET-CT effective dose of 10–12 mSv for whole-body imaging. For comparison, the mean effective dose of a diagnostic CT of the abdomen is 10 mSv ([48]). This radiation burden is 4–5 times the yearly natural background radiation dose according to the United Nations and is comparable with an equivalent of 500 chest radiographs ([49]). Exposure to radiation can be diminished by the usage of radioisotopes with a shorter half-life, which would be relevant in the case of repetitive scanning and as might be relevant for therapy monitoring.

This systematic review of mainly feasibility studies demonstrated that PET offers highly sensitive and potentially specific imaging of IA. For clinical practice, the technique may have value for the detection of early or residual subclinical synovitis. Specific tracers targeting inflammatory sites in the synovium are being developed and tested. Multicenter, large-scale validation studies with uniform, comparable outcome measures are obligatory to elucidate the definite role for PET in the early diagnosis and therapy monitoring of IA. One of the challenges is the development of standardized PET analysis methods. These studies could be performed in line with oncologic studies, for which a world-wide standardized guideline for tumor PET imaging and PET response criteria is available, allowing multicenter studies ([50, 51]). To further determine the strengths and weaknesses of PET as a diagnostic and monitoring tool for IA in relation to other imaging modalities, comparative studies in large cohorts are warranted. Validation studies should also address cost-effectiveness. Although the costs of PET are currently still high, they may decrease in time upon an increase in the applicability of PET in clinical practice. Moreover, PET may even save costs if it can be shown that PET can delay or most optimally prevent the onset of IA and/or increase the therapeutic efficacy of expensive biologic agents.

Finally, the hybrid imaging technique PET-MRI has been recently introduced in the first medical centers to use this technique. Preliminary PET-MRI fusion images of RA patients have shown interesting results ([28, 52]). PET-MRI combines highly sensitive functional imaging at the molecular level with superior anatomic details (including synovial tissue changes and bone edema) without the additional radiation burden of CT. In particular, the latter aspect may increase the options for practical implementation of PET for IA imaging. PET-MRI needs to be further explored in future studies of imaging of IA.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. van der Laken had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Bruijnen, Voskuyl, Hoekstra, van der Laken.

Acquisition of data. Bruijnen, Hoekstra.

Analysis and interpretation of data. Bruijnen, Gent, Voskuyl, Hoekstra, van der Laken.